Model-fitting approaches to the analysis of human behaviour.

Area of Science:

• Behavioral genetics
• Quantitative genetics
• Statistical modeling

Background:

• Model-fitting methods are increasingly used to analyze human behavioral variation.
• Diverse approaches exist for specifying these models, varying in their sensitivity to subtle sources of variation.

Purpose of the Study:

• To review and compare different model-fitting methods for analyzing human behavioral variation.
• To highlight the utility of maximum likelihood estimation for variance and covariance components.

Main Methods:

• Biometrical genetical approach for non-additive factors.
• Path analysis for environmental models with assortative mating.
• Maximum likelihood estimation for variance and covariance components.

Main Results:

• Different methods can yield similar conclusions, especially in simpler cases.
• Maximum likelihood is suitable for parameter estimation and hypothesis testing in multifactorial models.
• Experimental designs can reveal additive/non-additive genetic effects, mating systems, sibling/cultural effects, and gene-environment interactions.

Conclusions:

• Model-fitting, particularly with maximum likelihood, provides a flexible framework for analyzing complex behavioral variation.
• The choice of method depends on experimental design and data summarization efficiency.
• Extensions to multiple measurements and discontinuous traits are feasible.